Understanding Conditions of Uncertainty in Disease Surveillance and Response in Burkina Faso

(1)

Understanding Conditions of Uncertainty

in Disease Surveillance and Response in Burkina Faso

An inquiry into the sociological aspects of health information construction and use

Stine Loft Rasmussen

A thesis submitted in partial fulfilment of the requirements for the degree of Ph.D.

Department of Informatics

Faculty of Mathematics and Natural Sciences University of Oslo

March 2019

(2)

© Stine Loft Rasmussen, 2019

Series of dissertations submitted to the

Faculty of Mathematics and Natural Sciences, University of Oslo No. 2170

ISSN 1501-7710

reproduced or transmitted, in any form or by any means, without permission.

Cover: Hanne Baadsgaard Utigard.

Print production: Reprosentralen, University of Oslo.

(3)

Abstract

Through an interpretive, social constructivist approach, the research explores the production and use of information in the public health care domain of a low and middle-income country (LMIC). The research shows that taking action for health is as much about finding ways to deal with various forms of uncertainty or incomplete certainty as it is about producing better and more data.

Theoretically, the research considers the relationship between Information Systems (IS) and knowing. This is a long-standing yet contemporary theme in IS research and development as new enhancements in data-producing technologies stimulate an increasing demand for data globally, for example reflected in the debates around big data. Concepts from the domains of sociology of science and technology and from medical anthropology are drawn into the IS and HIS domain to develop a theoretical lens that takes into account the diverse contextual, situated and material character of how health workers and managers acquire knowledge and take action in the process of providing and administering care in the context of disease surveillance and response. These processes are conceptualized as multiple, cross-disciplinary, epistemic practices where uncertainty plays a key role.

Empirically, the thesis explores how information about the detection and transmission of two diseases (meningitis and dengue) are represented and acted upon in the context of disease surveillance and response in Burkina Faso. The ethnographically inspired case study of data use practices in the domain of disease surveillance and response in the health sector of Burkina Faso explores how health managers and workers create and use health information for the purpose of preventing and controlling contagious diseases. The study highlights how their work includes accounting for, rather than avoiding, different types of uncertainty.

The contributions of the thesis lie both in the sociological conceptualization of how knowledge about a disease is produced and used in the context of disease surveillance and response as well as in the articulation of different aspects of uncertainty and their relation to the understanding of how we know about a disease.

The articulation of uncertainty as not just a barrier to knowing but also a driver of action, helps supplement IS and HIS theory concerned with knowing. Uncertainty is introduced as

(4)

something that should be considered an important element to action rather than an element that must be avoided and eliminated through the formal IS or HIS. In this sense the findings invite us to flip the context for IS research and reflect further on what IS research can learn from LMICs instead of narrowly focusing on what we as IS researchers and practitioners can bring to this context.

(5)

Acknowledgements

During the past four years, I have been fortunate to experience the support and generosity of many people. I want to extend my gratitude to all of you. You have all taught me a lot in so many different ways, and this thesis is the sum of all these influences.

I am especially indebted to my supervisors Sundeep Sahay and Jørn Braa. Sundeep, you have been inspiring, critical and encouraging throughout the process. Your unwavering support – also through the last stretch – has been immensely valuable. Jørn, I have greatly appreciated your immediate trust and warm welcome into the HISP community. I also want to thank Jens Kaasbøl for getting me on track and Kristin Braa for helping me make some critical decisions and thinking of me even though I was sometimes far away.

This work would not have been possible without the hospitality of many of the hard-working people in the health sector in Burkina Faso. I cannot thank you enough for letting me unconditionally into your daily work and routines. Boureima Ouedraogo, thank you for opening the door to the Ministry of Health in Burkina Faso. Assetta Bara, thank you for accepting me and being my friend since the very first time we met. Boukary Ouedraogo, I feel fortunate to have shared the PhD writing experience with you and I am forever grateful that you showed me how to navigate an unfamiliar system. I also want to thank Tomé Ca and Albert Ouedraogo in WAHO for your inspiration and big smiles. Outside the Ministry of Health, I would like to thank Rebecca Greco Koné for helping me understand the many facets of disease surveillance and response.

I am grateful to all my colleagues both at the Department of Informatics and around the world who at different points in time have engaged in discussions or read and commented on my work. Margun Aanested, Hanne Cecilie Geirbo, Johan Ivar Sæbo, Petter Nielsen, Eric Monteiro, Terje Aksel Sanner, Troels Mønsted and Sisse Finken, your input has been very helpful. I also want to thank fellow students and friends for filling cold Norwegian winter days with warmth and stimulating discussions. Esther Namatovu Landen, Arunima Mukherjee, Wilfred Senyoni, Egil Øvrelid, Ayub Manya, Patrick Chikumba, Rangarirai Matavire, Selamawit Molla Fossum, Marta Vila, and Mette von Deden, thanks for sharing this experience and all the best of luck to all of you. From the wider HISP community I am happy to have met and worked with Edem Kossi, Zeferino Saugene, Jerry Aziawa, Romain Tohouri, Vincent Shaw, and Knut Staring.

(6)

I would like to thank Helle Samuelsen and Lise Rosendal Østergård from the University of Copenhagen for sharing your deep knowledge on Burkina Faso, and for always having time for a coffee and discussion.

Finally, I extend my gratitude to my family, who not only grew along with the thesis but also lovingly supported me through the strains and joys of the research process.

(7)

List of figures

Figure 1: HIS as Epistemic Machinery ... 43

Figure 2: Schematic overview of fieldwork ... 52

Figure 3: Organizational chart of the Ministry of Health. ... 53

Figure 4: Chains of translations in the meningitis surveillance system ... 79

Figure 5: Timeline of the 2016 outbreak of dengue in Ouagadougou ... 83

Figure 6: Analytical framework ... 91

(10)

Lists of tables and images

Table 1: Research sites ... 54

Table 2: Overview of interviews and full days of participant observation ... 56

Image 1: Configuration workshop on integrating nutritional data to DHIS2 ... 46

Image 2: Map of the health regions in Burkina Faso ... 55

(11)

Abbreviations

ANT Actor network theory CSF Cerebrospinal fluid

CSPS Centre de santé et promotion sociale (primary health facility) CSCW Computer-supported cooperative work

DHIS2 District health information system (version 2)

ENDOS Entrepôt national des données sanitaires (national health information data warehouse)

GIS Geographical information system HIS Health information system

HISP Health information system program

ICT Information and communication technology

ICT4D Information and communication technology for development IDSR Integrated disease surveillance and response

IHIA Integrated health information architecture IS Information system

LMIC Low and middle income country MoH Ministry of health

NTD Neglected tropical diseases RDT Rapid diagnostic test

SCOT Social construction of technology SST Social shaping of technology

TLOH Télégramme lettre officielle hebdomadaire (weekly epidemiological bulletin) WAHO West African health organization

WHO World health organization

(12)

(13)

Prologue

“Laafi?”

“Laafi bala”

This short sequence of words is exchanged several times a day in the Central Region of Burkina Faso. It is a greeting in Moré – the most widely spoken language in this region – and it is exchanged in a similar way as ‘How are you?’ ‘Fine thanks’ is exchanged in many Anglophone societies. In Moré, however, Laafi does not just mean fine, it means health. Being well thus has strong connotations of being healthy. While such a connection between health and well- being is probably true in most societies, it has a profounder meaning in a country such as Burkina Faso, where there is neither a free public health service nor affordable health insurance to rely on in cases of encounters with different diseases. For the ordinary citizen there is inherent uncertainty in accessing health care, whether it is through the public or the traditional system. For health workers and managers in the public health system there is inherent uncertainty in providing health care. They deliver health services under conditions of uncertain funding and constrained human, diagnostic and therapeutic capacity. Unexpected developments of potentially epidemic diseases and limitations in the tools to foresee and spot these developments generate further uncertainty.

When Health Information Systems (HIS) are introduced under these circumstances it is often with the aim of increasing the availability of information that can reduce uncertainty and improve health service delivery, for example identifying and tracking new cases of potentially epidemic diseases to support the making of data-driven decisions. However, the production and use of health data is often underpinned by imprecision, leading to partial uncertainty and in some cases complete uncertainty.

Through an ethnographically inspired case study of data use practices in the domain of disease surveillance and response in the health sector of Burkina Faso, this thesis considers how health managers and workers create and use health information for the purpose of preventing and

(14)

controlling contagious diseases. The study highlights how their work includes accounting for, rather than avoiding, different types of uncertainty.

From an HIS perspective I argue that these insights should inspire a shift of focus from the production of more and more precise data to considering how to accept uncertainty as part of decision-making processes, which in turn would raise the question of how to represent uncertainty in the HIS. The study suggests that this includes paying attention to both the multiplicity of diseases as well as the related practices of health information construction and use where other technologies such as telephones and Excel spreadsheets complement the electronic web-based HIS.

“Because in our country, if we have to wait until everything is confirmed, that will be after the death of many patients.”

(Health district manager, Ouagadougou)

(15)

Chapter 1 Introduction

In this section, I provide an outline of my research project on the sociological aspects of health information construction and use. The project considers the production and use of information in the public health care domain of a low and middle income country (LMIC).

Empirically, the thesis explores how information about the detection and transmission of two diseases is represented and acted upon in the context of disease surveillance and response in Burkina Faso. Theoretically, concepts from the domains of sociology of science and technology and from medical anthropology are drawn into the IS and HIS domain to develop theoretical concepts that take into account the diverse contextual, situated and material character of how health workers and managers acquire knowledge and take action in the process of providing and administering care. These processes are conceptualized as multiple cross-disciplinary epistemic practices. Taking action for health is shown to be as much about finding ways to deal with various forms of uncertainty or incomplete certainty as it is about producing better and more data.

1.1 Motivation, research aims and questions

Data has been in high demand during the past decades, and this demand is still growing. It is claimed, for example, that data has surpassed oil as the world’s most valuable resource (The Economist 2017). Such claims are partly nourished by technological developments within computing, the internet and social media, which have enabled data collection and processing at an unprecedented scale and pace. Other factors, however, are also at play when considering how data is put to effective use. One is the political will to govern by what can be measured, something that requires quantification of phenomena, including social ones. This trend has been promoted by the new public management agenda, exemplified in the saying

‘what gets measured gets done.’ Another enabling condition is a widespread confidence in data-driven decisions. Within the health care domain, this is exemplified by the use of rigorous clinical trials aimed at producing evidence about best clinical practices, making evidence- based medicine the de facto gold standard for making decisions about health improvements.

However, as Alexander and Lyytinen (2017) point out, it is all too easy to overestimate the value of big data as a reliable and objective resource for decision-making in organizations without taking into account the human capacity to process the data and make sense of it.

(16)

An important related factor concerns the process of knowing. The availability of data and techniques of data mining and machine learning has been celebrated as enabling new ways of knowing about the world that may lead to the ‘death’ of analytical professions such as epidemiology and financial analysis (Sahay 2016). However, it has also raised questions about the “avalanche of numbers” generated by these systems and whether these numbers provide clarity in knowing, or if they instead threaten to limit local initiatives needed for creating and maintaining good lives (Appadurai 2013, p. 299). In other words, the argument is that our ability to envisage better futures and facilitate social change does not rely solely on data and information. It is also linked to more fundamental human questions concerning how to handle uncertainty, which have been made almost invisible amid Western ideas about modernity and progress (Appadurai 2013). How we handle uncertainty, however, is deeply relevant when it comes to our ability to deal with life as it evolves rather than as it has been planned (Whyte 1998) and, more specifically, to doctors’ ability to diagnose and treat patients (Fox 1980;

Timmermans and Angell 2001).

Information Systems (IS) are vehicles for the collection, analysis and dissemination of data, making them intrinsic to discussions concerning data as a driver of change, and of knowing and doing (Orlikowski 2002; Suchman [1987] 2007). IS are often introduced with a promise to deliver more and more data to enable organizational and social change. However, the IS literature has provided many examples showing that this promise remains partially or fully unrealized (Sahay et al 2017) and that it can lead to other unanticipated changes. Such initiatives always carry tensions arising from the interplay between technology on the one hand and human and institutional practices on the other hand (Orlikowski 2007). This tension is magnified when data is seen as a commodity, or as an output in isolation from the enacted practices that produce it (Walsham 2001). However, despite these tensions, the quest for enhancing the production of data is continuous, in both research and practice. These tendencies are especially evident in the domain of HIS in the context of LMICs, the focus of this thesis, where resources are limited and data in high demand.

This thesis takes a deeper look at the dynamics of health information production and use in the context of disease surveillance in an LMIC. It considers the work that goes into the production of information about health and the way this information circulates and is used by health workers and managers at different levels of the health sector in their daily work. The

(17)

study conceptualizes these dynamics not as a consequence of information or information technology in itself, but instead as a social phenomenon comprising the interplay between technology and situated practices. Health information construction and use is enabled by the web-based HIS and other technologies, but it is the everyday practices involving both manual and computer-based work that shape how the information is used, for what purpose and by whom. Such a conceptualization requires a shift in focus from information as an input and output of the HIS to how a phenomenon (e.g. a disease under surveillance) is represented through the system, and how these representations lead to the construction of varying meanings to different actors. These varying meanings are conceptualized as ‘interpretive flexibilities’ and ‘multiplicity of representations.’ More specifically the thesis is about understanding how knowing about diseases is made possible in the health system of an LMIC, and how this knowing is intertwined with the practices which produce it and with those it informs. This broad approach is seen to reflect the sociological aspects of health information construction and use where the tools, technologies, materials and practices that go into producing knowledge and taking action hang together in a social environment.

The empirical setting for this theoretical development is the disease surveillance and response system in Burkina Faso. Disease surveillance and response is an area of vital importance to human well-being as it is concerned with preventing outbreaks of contagious diseases through early detection and response mechanisms. The work that goes into disease surveillance and response is heavily reliant on different types of information, ranging from early notification about potential cases to laboratory confirmed results detailing the more exact nature of the disease in question, and the resulting interventions based on this information. The devastating 2014-2015 Ebola outbreak in West Africa showed that disease surveillance and response is an area of global concern and that it is not only highly dependent on hard facts but also requires a functional health system that can support the interpretation of a plethora of information, including when it is inadequately confirmed or unconfirmed.

There is an ongoing degree of uncertainty surrounding the production and use of disease surveillance and response information. This uncertainty is particularly pertinent in LMICs, which operate under conditions of inadequate resources and infrastructure and frequent disease outbreaks. However, despite these uncertainties and inadequacies of available information, ‘the show must go on.’ People must be treated and epidemics controlled. This

(18)

thesis seeks first to understand how such practices unfold in context, and second to discuss how this invites us to think about how uncertainty can be better represented in IS and HIS design and research.

The thesis aims to contribute to discussions around the role of data in human action and social change. I conceptualize the relationship between the production of knowledge and its use for action in the domain of health service delivery in LMICs as a social-technical phenomenon that involves data, technology, people, and their physical and organizational settings. To do so, I take the view from sociology of science and science and technology studies, which emphasize that both knowledge and technologies are socially constructed (Bijker, Hughes, and Pinch [1987] 2012) but still represent reality (Latour 2005, 1999; K. D. Knorr Cetina 2009).

Specifically, I draw on the insight that knowledge about diseases can be constructed differently, with the implication that diseases are multiple as objects (Mol 2002; Moser and Law 2006). I furthermore combine these concepts with the notion that medicine is also about handling conditions of uncertainty (Whyte 1998; Cooper and Pratten 2014; Timmermans and Angell 2001; Fox 1980).

I have drawn these concepts together into a theoretical framework outlining the sociological aspects of health information construction and use. This framework acts as a theoretical vehicle for conceptualizing health information construction and use as a social process that does not inject a dichotomy between the mental and the physical worlds, but instead articulates the idea that what we can know – about a disease, for example – is intertwined with how we come to know in different settings and situations. Furthermore, the framework is designed to help us consider uncertainty not only as the absence of information, a gap that needs to be filled, but also as a driver that motivates us to ask further questions and thereby obtain more information. In other words, the theoretical framework proposed here assumes that obtaining information and knowledge about diseases is embedded in practices of handling uncertainty. The main research question is:

How can the concept of uncertainty inform our understanding of the challenges to health information construction and use in the context of low- and middle-income countries (LMICs)?

(19)

The concept is applied to an analysis of the empirical case of the disease surveillance and response system in Burkina Faso, and for this purpose the following empirical research questions have been applied:

 What are the networks of people, technology, and materials that enable knowing in disease surveillance and response?

 Which conditions of uncertainty affect production and use of information in disease surveillance and response?

Through the development and application of this framework for studying sociological aspects of health information construction and use, I hope both to provide a critique of the current data-centric approach to IS and HIS design and implementation and to outline an alternative to it. This study aims to provide a constructive critique that both deconstructs knowledge processes and offers inspiration for how to deal with the future as it is emerging in other ways than by numbers (Ortner 2016; Appadurai 2016).

1.2 Empirical setting and research approach

I have explored the research questions through an in-depth longitudinal study of health information construction and use in the health sector in Burkina Faso with a particular focus on its disease surveillance and response system. I have used an interpretive approach where the focus is on understanding why actors do what they do, and where the production of thick descriptions is emphasized as a tool to convey this understanding (Walsham 1995).

Methodologically, I have been inspired by ethnography, which has helped me to develop thick descriptions of the construction and use of health information in disease surveillance and response.

The fieldwork was conducted over a three-year period within the health sector in Burkina Faso. During this period, I lived in Burkina Faso and was affiliated with the IT directorate of the Ministry of Health as a ‘stagiaire’, which literally translates to intern, but in practice meant that I was considered a staff member of the directorate. From this position, I was able to do an in-depth interpretive study of work practices in the health sector in a West African country.

The research was done in affiliation with the Health Information Systems Programme (HISP) network. The HISP network is a research program working on strengthening health systems through the development of the free and open source software DHIS2, strengthening of public

(20)

health informatics capacity in LMICs as well as conducting research on HIS-related issues. In 2013 members of the HISP network assisted the Ministry of Health in Burkina Faso with the implementation of a national HIS database with DHIS2 as the core. Fieldwork for this thesis began in late 2014 after the system had been put in place, which means the research took place in the post-implementation phase.

Both the theoretical approach and the empirical methods are concerned with understanding how health managers and workers work with both information and absence of information. A key focus is on the social shaping of knowledge construction in the clinical setting as a means to understand both the content and processes of HIS. Theoretically, I conceptualize this as

‘epistemic machinery’ consisting of people, tools, and materials. The thesis aims to understand how this machinery works, including under conditions of uncertainty, which entails organizing based on both structure and improvisation.

1.3 Papers and research contributions

The papers that constitute the basis for this thesis are listed below. They all focus on lessons learned through a practice-based perspective on health information construction and use. The first three deal specifically with the case of disease surveillance and response, whereas the last paper has a broader focus on health data use and planning in general. Each paper deals with different aspects of the main thrust of the research: the construction and use of health information in disease surveillance and response in the health sector in Burkina Faso. The papers are listed below with an indication of their publication status.

1) Practices of disease surveillance and response in Burkina Faso

Rasmussen, SL (2017). In Proceedings of the 14th IFIP WG 9.4 International Conference on Social Implications of Computers in Developing Countries, Yogyakarta, Indonesia, May 22-24, 2017 (pp. 333–344). Springer, Cham.

2) Information and substance as defining elements of the epistemic cultures for meningitis surveillance and response in Burkina Faso

Rasmussen, SL and Sahay, S. Under review by The Information Society (status: revise and resubmit).

3) Disease surveillance and response in conditions of uncertainty: The case of ‘palu dengue’ in Burkina Faso

Rasmussen, SL and Sahay, S. Under review by Information and Organization (status:

revise and resubmit).

(21)

4) Plans and “off-plan activities”: exploring the roles of data and situated action in health planning in Burkina Faso

Rasmussen SL (2018). Electronic Journal of Information Systems in Developing Countries. 2018;e12049. https://doi.org/10.1002/isd2.12049RASMUSSEN 9of9

Paper one adopts a practice-lens to look at the characteristics of practices under the Integrated Disease Surveillance and Response (IDSR) and the use of their supporting guidelines. It shows that surveillance and response have two different sets of practices and there are challenges in keeping them connected. Surveillance is well grounded in practice and similar to routine HIS data collection, while response is often unexplored in practice and therefore more challenging to incorporate into existing HIS.

Paper two develops a sociological lens for health information construction and use in order to trace its epistemic practices, including the translations involved in producing representations of meningitis that support its surveillance and response. Meningitis is a critical disease to control in Burkina Faso, with relatively well established practices. The paper shows that the ability of the health staff to know relies on more than one epistemic practice, significantly defined by whether the material origin of the information is record or substance. Viewed in this perspective, integration of HIS becomes a challenge of integrating different epistemic practices.

Paper three explores the conditions that contribute to uncertainty in the case of an outbreak of dengue, a neglected tropical disease (NTD)¹, in Ouagadougou, the capital city of Burkina Faso. The analysis uses the perspective of multiple rationalities to discuss how what we do not know can be an asset to compensate for lack of prior experience, inadequate resources, ill- designed protocols and institutional constraints, which all contribute to creating uncertainty.

Health workers deal with these uncertainties by relying on improvisations such as making exclusions, exchanges through social relationships, and structuring practices informally. The

1 NTDs are a diverse group of diseases that persist in poor and marginalized communities although they have largely been eradicated elsewhere. They affect many people but measures to control them often remain unprioritized. Source: http://www.who.int/neglected_diseases/diseases/en/

This means they are often overlooked, and that when they appear it is under conditions of uncertainty.

(22)

paper argues that while the formal HIS does play a key role in conveying information, it is not well equipped for informing practices faced with conditions of uncertainty.

Paper four focuses on how health managers and workers use the concept of ‘off-plan activities’ to organize health activities in an uncertain context where local plans are often overruled at short notice by requirements from higher-level, unexpected developments in diseases, or sudden possibilities of funding. HIS data is conceptualized not as a tool to discipline local practices but rather as one of the elements that can guide individuals in their daily work.

As seen in these brief summaries, I have used slightly different theoretical lenses in each paper to bring out different aspects of the overall thesis, as each contributes a different facet of a sociological understanding of health information construction and use. Taken together, the papers constitute the empirical basis for the theorization of this perspective, which I elaborate on further in the thesis.

The findings I present in this study highlight that uncertainty is inherent to disease surveillance and response. In supplement to general IS and HIS theory it is shown that uncertainty is not only due to poor data quality or lack of information. Instead, it is seen as inherent in the conditions of health service delivery in general and in the production of knowledge about diseases as it is made visible first in the individual patient and later in the population. The findings also show that in LMICs, where resources are scarce, uncertainty is also seen in the lack of control over funds and health agendas. Furthermore, contrasting dengue and meningitis – two quite different diseases – shows that some types of uncertainty are more present in a disease such as dengue, which is not on global agendas, whereas with meningitis, which receives considerable attention from the international health community, these uncertainties are less prevalent when it comes to the construction and use of health information. On the ground, health managers and health workers creatively employ different practices to handle uncertainty. Concepts such as off-plan activities, and use of technologies such as mobile phones and spreadsheets, are used to make the data construction and use flexible so it supports everyday action by the health staff.

(23)

On the basis of these findings I outline the two main contributions of the thesis, which are:

 A sociological conceptualization of how knowledge about a disease is produced and used in the context of disease surveillance and response; and

 An articulation of different aspects of uncertainty in disease surveillance and response and their relation to the understanding of how we know about a disease.

The first contribution is mainly practical as it serves as a map of the different information requirements across the health sector in disease surveillance and response. It can be applied as a model to inform future HIS design aiming to strengthen disease surveillance and response.

It furthermore highlights that collaboration is not achieved through sharing the exact same information throughout the health sector, which lies at the heart of current integration efforts. A disease consists of multiple objects and their different representations through the HIS are situationally created and used in different ways by health managers and health workers.

The second contribution is theoretical as it describes variations in uncertainty and articulates it as not just a barrier to but also a driver of action. This articulation supplements IS and HIS theory concerned with knowing by introducing uncertainty as an element that should be considered important to action rather than as an element that must be avoided and eliminated through the formal HIS. This insight entails questions of how to represent uncertainty in HIS and IS.

Focusing on the sociological aspects of health information construction and use emphasizes the interplay between practices within different social conditions. It entails seeing this process of construction and use as embedded in everyday practices, as consisting of chains of translations, as concerning objects that are multiple but not fragmented, and as containing uncertainty that does not necessarily have to be avoided. It treats the link between data and action not as a linear sequence of practices but as a bundle of situated practices. This perspective has the advantage of showing that there is an order in practices even though they may appear random and untidy at first sight. It helps to explain that these practices can actually be quite well appropriated to enable action in conditions of uncertainty where knowing cannot be taken for granted.

(24)

1.4 Structure of the thesis The thesis is organized as follows:

In Chapter 2 on the context of the research, I position the research theme within different aspects of providing health data for action in LMICs. This extends HIS research in LMICs as it includes the health situation in Burkina Faso, global health agendas and recent efforts in disease surveillance and response globally and in West Africa.

In Chapter 3 on related literature I present the theoretical readings that I draw on to develop an understanding of the sociological aspects of health information construction and use. Next, I discuss social shaping as an approach to Information Systems, as well as the concept of multiplicity of representations. Finally, I introduce the idea of uncertainty leading to an ethics of possibility as an alternative to the ethics of probability.

In Chapter 4 on methods, I describe the methods used as being interpretive and related to the theoretical reading through a practice-based approach inspired by ethnographic work-place studies. I also provide an overview of the data collection and analysis processes.

In Chapter 5 on the analysis, I present a brief overview of how disease surveillance and response is done in Burkina Faso before I analyze two contrasting cases of disease: a neglected tropical disease and one that has recently received close attention from the global health community. In analyzing the material, I develop the two main ideas. The first is that epistemic practices are not shaped only by multiple rationalities but also by the multiplicity of the object that needs to be understood in order to be accounted for in an HIS. The second is the idea of uncertainty as a resource for the production of social order and aspirations for the future.

In Chapter 6 on contributions, I discuss how these findings contribute to current discussions in the HIS literature, including how they might be extended to the broader IS literature. This includes a discussion on the practical and theoretical implications of these insights as well as perspectives for future research.

Chapter 7 closes the thesis with some brief concluding remarks.

(25)

Chapter 2 Context of the research

This chapter presents the context of the research and the challenges with health information construction and use that it aims to address. I begin by introducing the site where the fieldwork was undertaken, namely the health sector in Burkina Faso, with a focus on the current HIS situation there. Next, I broaden this focus to discuss HIS in developing countries more generally with a particular focus on the development of the free and open source software DHIS2 by the HISP network, which is the platform used for the national HIS in Burkina Faso. I end the chapter with a view on the broader context of global health through a brief discussion of some current trends in this field, underlining that there are different expectations around the HIS in general and disease surveillance and response in particular.

2.1 Burkina Faso: uncertainty and hope

Despite some recent turmoil, Burkina Faso is a politically stable country. The former president, Blaise Compaoré, managed to steer the country free of other conflicts in the wider West African region. In 2014 his 27-year rule came to an end due to a successful popular uprising motivated by his intent to change the constitution in order for him to serve yet another term.

The uprising led to his resignation and ultimately peaceful democratic elections a year later in 2016. This change in the political landscape has spurred optimism on one hand, as it has become a pan-African example of the power of populations to drive political change (Hagberg 2015). The new government, however, has been challenged as the political instability and insecurity in the neighboring country of Mali has started to spill into Burkina Faso. Areas along the Malian border particularly are affected by small but frequent extremist attacks on public services such as police stations and schools, leading to great insecurity and uncertainty among the population in these areas about their future. The situation has led to a large number of internally displaced people. The capital of Ouagadougou has seen three large terror attacks in the first two and a half years that the new government has been in power. As two of these attacks were directly targeting Westerners, they have affected tourism and foreign investments adversely, leading to loss of employment and business opportunities. Taken together these events contribute to general uncertainty about the future, lined with both hope and worry.

(26)

2.1.1 Public health

With up to 40 percent of the population living below both national and international poverty lines Burkina Faso is among the ten lowest ranked countries in the UNDP Human Development report (UNDP 2016). The human resources for health are quite limited; for example there is one medical doctor per 15,836 inhabitants (Ministère de la Santé Burkina Faso 2016). In contrast there are between three and five medical doctors per 1,000 inhabitants in most European countries (WHO 2017). The health sector is highly donor dependent, which creates a culture of planning that on the one hand targets the most urgent health needs, but on the other hand is constantly being reshaped by these external funding mechanisms. The extensive external funding entails quite some coordination and negotiation when it comes to financing and delivering health services.

In general, public health systems in developing countries are characterized by uncertainty as living conditions and delivery of health services are influenced by poverty, risk of armed conflict, variability of infrastructure, and weak governments and public sectors (Farmer et al.

2013). Burkina Faso is not an exception. Health workers have to deal with multiple contingencies, not just those directly related to care provision, but also in emergency situations where measures to protect health workers are not in place, as was the case during the 2014-2016 Ebola outbreak that unfolded in countries close by (Østergaard 2015). Patients are seldom sure that health clinics can deliver what they need, and “caring for a sick child becomes a routine based on uncertainty” (Østergaard, Bjertrup, and Samuelsen 2016). Often health seeking behavior entails a multitude of strategies of swapping between the formal and traditional components of public health care delivery (Samuelsen 2004).

One health area that has gained significant attention following recent large-scale epidemics in the West African region is on strengthening disease surveillance and response capacity of the health system. Located in the middle of West Africa, Burkina Faso has borders with six other countries, making it vulnerable to communicable diseases that are easily spread over the borders. This threat is enhanced by the fact that there is substantial migration aided by extensive travel between the francophone West African nations (in ECOWAS). Burkina Faso has responded to these threats by adopting both the international Integrated Disease Surveillance and Response (IDSR) guidelines as well as being committed to the Global Health Security Agenda, emphasizing the ‘one health’ approach (CDC 2016). These commitments

(27)

have led to a number of national and regional initiatives, including some to strengthen the national HIS capacity to support disease surveillance and response.

2.1.2 Health system, national HIS and DHIS2

The health sector in Burkina Faso is organized at three levels: national, regional and district.

The national level is in charge of strategic development of the health sector, including the development of clinical guidelines and coordination of national health initiatives such as vaccination campaigns. The regional level is an administrative level that oversees and assists districts in health service delivery. The district level is operational and is in charge of managing health care delivery through the health facilities in the district. This includes implementation and local coordination of national initiatives, monitoring of quality of services, and administrating drugs and medical equipment. In 2016 there were 13 regions, 70 districts, and approximately 1,760 health facilities (CSPSs) across the country, each serving on average 10,000 patients (Ministère de la Santé Burkina Faso 2016). There are a number of hospital-like clinics, as well as regional and national level hospitals, but the CSPSs are the primary entry point to the health sector for most of the population. There are some private clinics but mainly in the capital of Ouagadougou, where more people can afford to pay for health services.

Laboratories share premises with a clinic or a district but in some cases they are managed independently.

The national HIS in Burkina was strengthened in 2010-2013 through the development of a strategic HIS plan and the installation of the DHIS2 platform for the collection of routine national data. This work was assisted by a team from the HISP program at the University of Oslo (Braa and Sahay 2012). Since 2013 the District Health Information System software version 2 (DHIS2, https://www.dhis2.org/) has been in use as the backbone of the national HIS in Burkina Faso, including for the IDSR programme. The Ministry of Health is now working with limited support from the HISP network with maintenance, redesign and adaptation of the system. In terms of content the database covers most of the national health programs when it comes to routine health information collection. Data entry is done at health district level, as there are no computers at the health clinics. There are data managers employed at the district, regional and national levels, but not at the CSPSs.

(28)

2.2 HIS in LMICs and the HISP network

The research is part of the global Health Information System Program (HISP), which is engaged in strengthening of HIS in LMICs (Sahay, Sundararaman, and Braa 2017; Braa and Sahay 2012;

Braa, Monteiro, and Sahay 2004). Over a 20-year period, HISP has gone from being a small- scale, bottom-up development project initiated in South Africa to becoming a large global player encompassing a vast organizational network delivering and implementing a software program that supports routine national registration of public health data in more than 70 countries. This expansion and globalization means that the HISP network is now also engaging in more top-down global agendas of data-driven development and standardization. While the expansion is due to the ability of the program to solve many challenges concerning the quality of HIS, new challenges are also arising due to the scale of the program and the ensuing complexity (Sahay, Sundararaman, and Braa 2017). In this thesis, I focus on one of those emerging challenges: the one of health information construction and use in the context of disease surveillance and response in Burkina Faso.

2.2.1 Creating a better world (with information for action)

The HISP program adopts a comprehensive approach to information systems. There are three core components: action and research around the design and development of the free and open source software DHIS2 (see dhis2.org for more details), the promotion of an integrated health information architecture (IHIA), and capacity building on public health informatics in countries. The DHIS2 software is designed specifically to serve the district-based health systems of many LMICs, while the IHIA promotes a way to streamline the fragmented landscape of databases, lab systems and electronic health records in LMICs (Braa and Sahay 2012). The capacity building in public health informatics is done through various means including hands-on training in use, implementation and administration of the DHIS2 software, enrolment of PhD and Master’s students, and continuous action research activities.

In recent years, HIS has specifically been adopted in LMICs in order to support the provision of health information to facilitate health systems strengthening. This increase in adoption has on the one hand been made possible through technological advancements such as cloud computing and mobile phones (Adaletey, Poppe, and Braa 2013; Asangansi and Braa 2010).

At the same time, it has been aligned with a global demand for more data and indicators, creating a strategic and financial push to strengthen health data collection as a key element

(29)

in improving health systems (AbouZahr and Boerma 2005; Lippeveld 2001; Sahay, Sundararaman, and Braa 2017).

It has always been central to the work of HISP that the health information collected through the DHIS2 software should benefit countries by empowering health managers at all levels to make better decisions in their endeavors to improve health. An early slogan of the HISP was

‘information for action,’ which has also been reflected in partnerships with for instance the Health Metrics Network, which advocates HIS strengthening under the slogan ‘better information, better decisions, better health.’ This slogan gained global momentum with the introduction of the Millennium Development Goals and now the Sustainable Development Goals as the idea behind these agendas was to create globally aligned indicators against common goals for the world and enable the measurement of development progress.

However, as the introduction of HIS expands beyond the pilot stage in many countries, a paradox is emerging: even as these HIS are becoming more stable and providing more comprehensive and timely data on a national scale, the use of this data remains limited (Sahay, Sundararaman, and Braa 2017). This is especially the case when it comes to the peripheral levels of the health sector, where significant amounts of time are spent on data collection while feedback is seldom, if ever, given.

Within the HIS literature, traditional explanations are given for low data use which are prone to assuming a rational, linear relationship between data and its use (Braa and Sahay 2012).

The idea behind this approach can be summed up in the illustrative but not so flattering saying

‘garbage in, garbage out,’ meaning that if low quality data is entered into an HIS, only low quality information can be extracted. This dynamic leads to a vicious circle where poor data leads to lack of trust in the information, which leads to lack of use, which in turn lowers the incentives to collect the data and so forth (Sahay, Sundararaman, and Braa 2017). Such types of explanation are sound and they do address part of the challenge with low data use.

However, they still imply that if data is collected, it will eventually be analyzed and used as a basis for action. This precludes an understanding of how people act under conditions of uncertainty.

Until recently, these challenges were mainly approached in a rationalistic way that focused on the provision of better and more data as the main key to improve use of health information.

(30)

However, the provision of useful information involves being more specific about who needs what information, under which circumstances and for what purposes (Sahay, Sundararaman, and Braa 2017). For example, it has been pointed out that taking action based on an HIS requires translating data into relevant knowledge (Lewis and Sahay 2009). Further efforts to reposition the formulation of data use in the health domain include seeing decisions in health as based on conversations that matter, rather than on hard evidence only (Kelly, Noonan, and Sahay 2013 ). This idea was recently explored in a paper that drew a distinction between data use practice being ‘authoritarian-bureaucratic,’ on the one hand, and ‘dialogic’ on the other (Kelly and Noonan 2017). In all these examples data use is closely linked to the kind of practices that go beyond mere production of data.

By being part of the HISP research initiative, this research project stands on the shoulders of much of the previous and ongoing HISP research, while it also engages in one of the new key topics, information use, to help drive the movement towards its goals of contributing to better health for the populations in LMICs. It is a move that requires a turn to public health informatics, which Sahay, Sundararaman and Braa (2017) describe as an emerging paradigm in the public health domain within the context of public health.

2.3 Global health and disease surveillance and response

Global health can be seen as a collection of health problems, addressed through the development and delivery of technologies (Farmer et al. 2013). It is more an agenda than a field, as its main focus is to engage in solving global health challenges through interdisciplinary efforts rather than through one discipline only. In this sense, it is similar to public health, although it differs in that it addresses both prevention and clinical care provision to individuals while public health mainly addresses health in the population (Koplan et al. 2009). Within global health it is recognized that it is important to provide not only clinical solutions to worldwide health challenges, but also solutions that address structural inequalities, poverty and unequal access to health care. The reasons that health services are not universally accessible to populations in countries around the world are often related to poverty and structural or economic constraints. For example, who should invest in developing medicines for people who cannot afford to pay for them? How can poor states finance the delivery of health services that are accessible for all regardless of socio-economic status? These are some

(31)

of the questions that underpin global health efforts, and they require both technical and structural solutions, including those concerning their HIS.

The year of 1978 was a milestone for global health. All countries of the world agreed on the Alma-Ata declaration, which was the first unified statement to secure health for all as a fundamental human right. However, the vision behind the Alma-Ata declaration suffered during the 80s and 90s as a result of financial crises and structural adjustment policies promoted by the IMF and the World Bank, leaving limited space for the investments needed to promote a ‘health for all’ agenda (Farmer et al. 2013). The politics of these years resulted in weak health systems that did not have the capacity to provide the services needed to achieve health for all. International organizations thus had to develop alternative interventions at lowered costs that were easy to implement and that could demonstrate effective results. United Nations International Children's Emergency Fund (UNICEF), for example, successfully employed selective primary health care on a large scale to fight child mortality and malnutrition through delivery of simple measures such as growth monitoring, oral rehydration treatment, breastfeeding and immunization (also referred to as GOBI). The results on global child mortality were impressive and the strategy functioned in spite of health systems being unable to deliver all basic services. However, financial support eventually dried up and exposed the weaknesses of this approach. Without funding, it was not sustainable as it did not target strengthening the health system in itself. Similarly, a new turn in health service delivery in LMICs is the involvement of large global nongovernmental organizations (NGOs) that exclusively target a specific disease. This yields impressive results in the targeted disease, such as HIV/AIDS, but again neglects to strengthen the overall health system in the country and as a result, standards of treatment remain low for other diseases, such as NTDs and respiratory illnesses. As it favors projects at the expense of a holistic sectoral approach, this approach has been labelled projectification of health services (Prince 2013; Meinert and Whyte 2014).

To the projects and donors who measure targets focusing on the diseases they address, this might not matter so much, but to the people who seek health services, it can be difficult to understand why there is treatment for some diseases while not for others. This is illustratively summarized in the statement ‘I wish I had AIDS’ made by groups of patients in Uganda (Whyte

(32)

2012). AIDS was at the forefront of the then global health agenda and received considerable funding and attention.

This approach leaves local doctors with a range of ethical dilemmas around where to work and how to treat people equally when the means to do so is not equal (Prince and Otieno 2014). Another dimension to take into account when it comes to data for global health is the purpose of the data collection. Crane has observed how HIV/AIDS treatment in Uganda has been transformed over a period of ten years from a “local clinic to a global research site”

(2013, p. 181) feeding new types of data on clinical trials to a global community of researchers.

Along these lines it has thus been argued that global health has become framed as a series of technical problems that need technical solutions, which furthermore marks a shift away from the public health priorities of securing health service delivery for everyone (Crane 2013; Prince 2013).

The potential consequences of the lack of focus on health sector strengthening became evident in the wake of the 2014-2015 Ebola epidemic in West Africa. The big question emerging from this event was how it could develop to such an unprecedented scale when outbreaks of the same disease in Central and Eastern Africa were usually quickly contained.

One key explanation was to be found in the weak capacity of the health systems of many West African countries as they had been severely underfunded for many years (Gostin and Friedman 2015).

Providing health information in the context of global health is a matter of balancing multiple interests and agendas (Sullivan 2017). In particular, there is a tension between, on the one hand, generating data on projects that easily show results because they are easy to measure, and on the other hand generating data to monitor long-term improvements in the health system itself. These are more difficult to measure and the definition of indicators is not as straightforward. For example, it is uncontested that high quality research data is needed for the development of better technology and medicines to improve health globally, but doctors and nurses across Africa are working hard to make ends meet at clinics and hospitals, and global agendas also need to prioritize improving conditions for their work.

(33)

2.3.1 Information for disease surveillance and response

A theme central to contemporary global health priorities is the prevention of communicable diseases and especially those at risk of spreading across borders and becoming pandemic. It is a theme that recurs frequently on international agendas (Zacher 1999). Large outbreaks of lethal diseases have been known throughout history. Due to advancements in medical science during the 19^th century a trust in medication was summarized by the slogan ‘one drug for each bug,’ and attention to disease surveillance waned more generally until the 1990s (Zacher 1999). Around this time there were large outbreaks of known diseases, such as cholera, but the appearance of new threats such as HIV/AIDS also sparked tremendous worries about the potential for uncontrollable outbreaks on a global scale, resulting in a renewed focus on disease surveillance and response (Zacher 1999). As mobility across and within borders continues to increase and new diseases continue to develop, it is clear that these diseases cannot be contained through medication and restrictive measures on movement only.

Instead, rapid detection and efficient deployment of control mechanisms is now seen as the key to cope with these new threats (Zacher 1999; Lakoff 2017).

In recognition of these challenges the member states of the World Health Organization’s African branch (WHO AFRO) adopted a resolution on IDSR in 1998 (CDC 2015; Kasolo, Roungou, and Perry 2010). The main idea behind the IDSR framework was to transform disease surveillance into a cross-sectoral undertaking instead of a task of each individual disease program, which had proven ineffective. This was envisaged through the standardization and streamlining of the technical guidelines down to each individual disease.

There are two kinds of guidelines in the IDSR framework: general ones and individual instructions for more than 45 diseases. Further, when a country adopts the standard IDSR guidelines, they also contextualize them by making their own country-specific guidelines, including identifying which particular diseases to include. In Burkina Faso, the guidelines were adopted in collaboration with the WHO.

The IDSR framework has quite a number of objectives (Kasolo, Roungou, and Perry 2010).

Those of most interest when it comes to supporting IDSR through HIS are to “integrate multiple surveillance systems,” “improve the use of information” and “improve the flow of surveillance information.” (Kasolo, Roungou, and Perry 2010, pp. 7-8). Doing so involves improving different data streams and acquiring surveillance data for early tracking of new

(34)

diseases. Improving the use of information implies generating relevant evidence to strengthen action for both surveillance and response.

Surveillance has to do with collection of information while response is about taking action, including being prepared to fight disease outbreaks. Surveillance refers to all the mechanisms put in place to detect the first cases of contagious diseases. This involves diagnosing and identifying potential cases as well as enabling health workers to diagnose – or at least know when to suspect – such cases and refer them appropriately to relevant health facilities. This could be about both putting in place rapid diagnostic tests and improving laboratory capacity.

However, for some diseases such tests are not easily available and diagnosing might instead be done clinically, supported by diagnostic guidelines. Surveillance also involves raising awareness in the population so that people know when to seek medical assistance. In any case surveillance also involves the rapid and efficient dissemination of information about suspected cases of potentially epidemic diseases through the health system.

Response is about all the different types of action that can be taken based on the information about an emerging outbreak or a new threat. This also involves effective control mechanisms of verification, so that it can be confirmed whether the suspected case is indeed a real case.

There are several such mechanisms in place in the health system but the final controlling body should be an independent organization, as countries might have too many economic interests clouding their decision on whether an emerging outbreak should be made publicly known or not (Zacher 1999). However, the response time of international agencies has also been hotly debated, with criticism for being both too slow to announce and too quick on the trigger (Lakoff 2017).

Again the 2014-2016 Ebola epidemic in West Africa has highlighted all too well that although the IDSR guidelines have been adopted in many African countries, there is still a need to improve disease surveillance and response efforts in this region (Phalkey et al. 2015). Some of the information related challenges that were identified after the Ebola epidemic were difficulties to diagnose fast enough, to provide up to date and accurate information to the highest levels, to coordinate among the many actors involved in the response, and to inform populations about the protective measures to take (Dahl et al. 2016; Gostin and Friedman

(35)

2015; Moon et al. 2015). In addition, the data on the effectiveness of the response was not sufficiently reliable to support further action (Cancedda et al. 2016).

In short, disease surveillance and response is about monitoring communicable diseases in order to identify new cases and controlling the transmission of these diseases before they spread and develop into epidemics. Ideally, this should generate information that can be used for public health action (Kasolo, Roungou, and Perry 2010). Interestingly, this is very close to the vision of ‘information for action’ that tends to underpin the HISP agenda around HIS design, and supposedly there should be mutual benefits in supporting IDSR data collection through established countrywide HIS. Frameworks such as the IDSR guidelines are providing standards and high-level descriptions of how work should be done. While on the one hand the guidelines are a gift for efforts to strengthen HIS, on the other hand they bring some challenges, especially concerning their adaptation to existing conditions in individual countries. They still need to be contextualized, and the guidelines are silent on how this contextualization should take place.

2.4 Health information for action: a challenge of knowing and doing

Based on the readings in this section, the challenge to health information construction and use appears to be a practical, contemporary challenge to the public health domain. However, the readings show that there are further challenges. These concern determining how information leads to action, as well as how to produce information for action that is relevant for stakeholders across the spectrum in the health sector. As HIS are tools to provide relevant information, they are inevitably implicated in such work. However, there are significant challenges when it comes to supporting meaningful data use beyond just routine reporting.

This question has mainly been approached as a data quality issue; in other words, the assumption is that better data would lead to more action. On the other hand, I argue that the question of health information use for action is also a question of how to collaborate in multidisciplinary settings where the character of action is diverse, which affects the nature of information required by different stakeholder groups. Furthermore, the tools to produce information are constrained. It is a challenge of knowing and doing under these circumstances as much as it is of data production. Such topics of knowing and doing have been of interest to the IS field, and it is with this in mind that I now turn to a discussion on how to approach these challenges theoretically.

(36)

Understanding Conditions of Uncertainty in Disease Surveillance and Response in Burkina Faso